煤资源在现代工业中应用广泛，然而近几年随着对煤资源大规模开采，煤矿中各类事故也时有发生。煤自燃是煤矿中最常见的事故，因其经常发生在危险的区域内，导致现场的煤自燃程度很难被及时掌握，从而无法做出相关的预防性处理。因此，众多研究机构通过对现场煤自燃发生的条件进行模拟，分析煤自燃过程中气体浓度与温度的变化情况，以期获得其相关关系，并实现对煤自燃温度的准确预测。

针对煤矿中煤自燃温度与气体浓度之间复杂的非线性关系，由于传统机器学习算法存在预测误差大、单一核函数的泛化能力弱和无法对异常数值处理等不足，本文提出一种基于模糊组合核相关向量机的煤自燃温度预测模型。首先，搭建模拟煤自然发火平台，通过实验获取不同气体浓度与煤自燃温度数据；然后，对气体浓度数据进行归一化处理，并用组合核函数构造输入样本矩阵，将低维空间的数据映射到高维空间，以获得更佳的训练模型；接着，用模糊算法赋予训练样本不同的隶属度，即不同的重视程度，以减小异常值对模型的影响；最后，构造组合核相关向量机模型对煤自燃温度进行预测，并与基于径向基神经网络、最小二乘支持向量机、高斯核相关向量机、组合核相关向量机的煤自燃温度预测方法做比较，验证模糊组合核相关向量机模型在煤自燃温度预测方面的有效性。

结果表明，模糊组合核相关向量机用于煤自燃温度预测不仅有泛化能力强、减弱异常数据的影响等特点，而且模型更稀疏，因此模糊组合核相关向量机在煤自燃温度等复杂问题预测方面具有明显的优势。

Coal resources are widely used in modern industry. However, with the large-scale exploitation of coal resources in recent years, various accidents in coal mines have also occurred from time to time. Coal spontaneous combustion is the most common accident in coal mines. Because it often occurs in dangerous areas, it is difficult to grasp the degree of coal spontaneous combustion at the scene in time and preventive measures cannot be taken. Therefore, many research institutions simulate the conditions of coal spontaneous combustion on site, and analyze the changes of gas concentration and temperature during the process of coal spontaneous combustion, in order to obtain the correlation and realize the accurate prediction of coal spontaneous combustion temperature.

In view of the complex non-linear relationship between coal spontaneous combustion temperature and gas concentration in coal mines, due to the large prediction errors of traditional machine learning algorithms, the weak generalization ability of a single kernel function, and the inability to deal with abnormal values, this paper proposes a prediction model of coal spontaneous combustion temperature based on fuzzy combined kernel relevance vector machine. Firstly, built a simulated coal spontaneous combustion platform, though experiment to obtain data of different gas concentrations and coal spontaneous combustion temperature; secondly, normalize the gas concentration data, and use the combined kernel function to construct the input sample matrix to map the low-dimensional space data to the high-dimensional space in order to obtain a better training model; then, the fuzzy algorithm is used to give the training samples different degrees of membership, that is, different degrees of importance, to reduce the influence of outliers on the model; finally, a combined kernel relevance vector machine model is constructed. The coal spontaneous combustion temperature is predicted and compared with the coal spontaneous combustion temperature prediction method based on radial basis neural network, least square support vector machine, Gaussian kernel relevance vector machine and combined kernel relevance vector machine to verify the fuzzy combined kernel relevance vector machine model validity in predicting the spontaneous combustion temperature of coal.

The results show that the fuzzy combined kernel relevance vector machine used in coal spontaneous combustion temperature prediction not only has the characteristics of strong generalization ability and weakened the influence of abnormal data, but also the model is sparser. Therefore, the fuzzy combined kernel relevance vector machine is used in the prediction of complex problems such as coal spontaneous combustion temperature. There are obvious advantages in this regard.

[1] 朱大庆. 我国煤、电厂商的短期市场势力与煤电一体化关系分析[J]. 北京交通大学学报 (社会科学版), 2018, 17(3): 72-81.

[2] 宁小亮. 煤与瓦斯突出预警技术研究现状及发展趋势[J]. 工矿自动化, 2019, 45(8): 25-31+37.

[3] 吕媛, 牛艳青, 雷雨, 等. 生物质与煤混燃灰熔融特性与动力学研究[J]. 工程热物理学报, 2020, 41(9): 2285-2290.

[4] 中华人民共和国国家统计局. 2020年全国原煤产量情况[EB/OL], http://www.stats.gov.cn/tjsj/ndsj/2020/indexch.htm

[5] 籍永华, 秦丙克, 黄沈丹. 汪家寨煤矿煤中汞的含量分布与赋存状态[J]. 煤炭技术, 2019, 38(7): 156-159.

[6] 王兴兆, 李传, 邓文安, 等. 煤溶胀对煤/油加氢共炼中煤转化率的影响及机理研究[J]. 煤炭转化, 2019, 42(2): 1-7.

[7] 黎经雷, 牛会永, 鲁义, 等. 风速对近距离煤层采空区漏风及煤自燃影响研究[J]. 煤炭科学技术, 2019, 47(3): 156-162.

[8] 梁运涛, 侯贤军, 罗海珠, 等. 我国煤矿火灾防治现状及发展对策[J]. 煤炭科学技术, 2016, 44(6): 1-6+13.

[9] Y. H. Zhang, P. L. Zhou, Z. A. Huang, et al. Yellow mud/gel composites for preventing coal spontaneous combustion[C]//Proceedings of 2015 International Conference on Materials, Environmental and Biological Engineering (MEBE 2015), 2015: 753-759.

[10] Y. Sun, S. G. Wang, L. Wei, et al. Coal spontaneous combustion characteristics based on constant temperature difference guidance method[J]. Process Safety and Environmental Protection, 2019, 131: 223-234.

[11] T. Xu. Heat effect of the oxygen-containing functional groups in coal during spontaneous combustion processes. 2017, 28(8): 1841-1848.

[12] M. W. Xu, L. Z. Jin, L. Yu. Selection of different gas indexes for spontaneous combustion of “three zones” in goaf[C]//Proceedings of the 7th International Symposium on Project Management (ISPM 2019), 2019, 1010-1016.

[13] 李晓洁. 煤与生物质加压热解及协同作用研究[D]. 杭州: 浙江大学, 2019.

[14] 李林, 陈军朝, 姜德义, 等. 煤自燃全过程高温区域及指标气体时空变化实验研究[J]. 煤炭学报, 2016, 41(2): 444-450.

[15] 朱红青, 沈静, 张亚光. 升温速率和氧浓度对煤表观活化能的影响[J]. 煤炭科学技术, 2015, 43(11): 49-53+106.

[16] Y. J. Zhang, G. G. Wu, H. F. Xu, et al. Prediction of oxygen concentration and temperature distribution in loose coal based on BP neural network[J]. Mining Science and Technology (China), 2009, 19(2): 216-219.

[17] 邓军, 周少柳, 马砺, 等. 基于PCA-PSOSVM的煤自燃程度预测研究[J]. 矿业安全与环保, 2016, 43(5): 27-31.

[18] 刘宝, 穆坤, 叶飞, 等. 基于相关向量机的煤自燃预测方法[J]. 工矿自动化, 2020, 46(9): 104-108.

[19] 焦振华, 林卫国, 黄洪涛, 等. 沿采空区不规则回采工作面分区瓦斯治理研究[J]. 煤炭工程, 2021, 53(1): 89-93.

[20] 王宪光, 陈志平, 龚邦军. 高瓦斯易自燃综放面采空区内因火灾防控[J]. 科技创新与应用, 2021(10): 146-148+151.

[21] S. Pramanik, A. Hussain. Text normalization using memory augmented neural networks[J]. Speech Communication, 2019, 109: 15-23.

[22] 张滕. 基于贝叶斯蒙特卡洛方法的风力发电机可靠性建模[D]. 天津: 天津大学, 2017.

[23] 范磊, 卫志农, 李慧杰, 等. 基于变分模态分解和蝙蝠算法-相关向量机的短期风速区间预测[J]. 电力自动化设备, 2017, 37(1): 93-100.

[24] H. Abbas, J. Tezcan. Relevance vector machines modeling of nonstationary ground motion coherency[J]. Soil Dynamics and Earthquake Engineering, 2019, 120: 262-272.

[25] D. D. Kong, Y. J. Chen, N. Li, et al. Relevance vector machine for tool wear prediction[J]. Mechanical Systems and Signal Processing, 2019, 127: 573-594.

[26] 刘恩旭. 基于优化的RVM铁路货车轴承温度状态预测的研究及实现[D]. 北京: 北京交通大学, 2019.

[27] Y. Yu, M. H. Su, J. L. Ma, et al. Corrosion acoustic emission signal recognition based on relevance vector machine[C]//Proceedings of 2017 13th IEEE International Conference on Electronic Measurement & Instruments (ICEMI), 2017: 1060-1065.

[28] J. Lei, Y. Y. Tao, X. Y. Su. Incremental learning algorithm based on relevance vector machine[C]//Proceedings of 2018 the 2nd International Conference on Advances in Image Processing (ICAIP 2018), 2018: 227-230.

[29] 王春雷, 赵琦, 秦孝丽, 等. 基于改进相关向量机的锂电池寿命预测方法[J]. 北京航空航天大学学报, 2018, 44(9): 1998-2003.

[30] 张旭, 张裕, 郑正奇, 等. 基于核函数与卡尔曼滤波的室内定位方法[J]. 上海理工大学学报, 2016, 38(3): 287-292.

[31] N. Zhao, X. Zhan, Y. T. Huang, et al. Kernel machine methods for integrative analysis of genome‐wide methylation and genotyping studies[J]. Genetic Epidemiology, 2018, 42(2): 156-167.

[32] Z. D. Tian, S. J. Li, Y. H. Wang, et al. Wind power prediction method based on hybrid kernel function support vector machine[J]. Wind Engineering, 2018, 42(3): 252-264.

[33] 张晓宁. 基于可变高斯核函数的最优分位数回归问题研究[D]. 杭州: 浙江大学, 2018.

[34] Y. B. Li, C. Liu, S. Y. Zhang, et al. Reproducing polynomial kernel extreme learning machine[J]. Journal of Advanced Computational Intelligence and Intelligent Informatics, 2017, 21(5): 795-802.

[35] 田萌. 高斯核参数选择与正交多项式核构造[D]. 太原: 山西大学, 2017.

[36] 周有荣, 王凯. 改进鲸鱼算法优化混合核支持向量机在径流预测中的应用[J]. 中国农村水利水电, 2020(7): 50-53.

[37] 刘昕玥, 姚建国, 万定生. 基于组合核函数SVM的中小流域流量预测研究[J]. 计算机仿真, 2019, 36(11): 454-457.

[38] 刘洪于. 基于模糊理论的导弹贮存使用环境分析[J]. 装备环境工程, 2020, 17(12): 95-100.

[39] 彭龙, 党三磊, 纪伊琳, 等. 基于模糊理论的计量检定设备健康评价模型[J]. 工业计量, 2020, 30(6): 90-92.

[40] C. G. Yin, H. X. Wang, H. Feng, et al. The set of fuzzy time series forecasting models based on the ordered difference rate[C]//Proceedings of 2017 2nd International Conference on Modelling, Simulation and Applied Mathematics (MSAM 2017), 2017: 44-47.

[41] B. Yang. Dynamic risk identification safety model based on fuzzy support vector machine and immune optimization algorithm[J]. Safety Science, 2019, 118: 205-211.

[42] 黄昕宇, 张栋良, 李帅位. 基于改进模糊支持向量机的汽轮机热耗率预测模型[J]. 热力发电, 2019, 48(3): 22-27.

[43] A. Rahbarzare, M. Azadi. Improving prediction of soil liquefaction using hybrid optimization algorithms and a fuzzy support vector machine[J]. Bulletin of Engineering Geology and the Environment, 2019, 78(7): 4977-4987.

[44] Y. Y. Ji, F. Ding. Application of fuzzy PID adaptive algorithm based on RBF network in smart car[C]//Proceedings of 2019 International Conference on Computer Science, Communications and Multimedia Engineering, 2019: 264-268.

[45] 王振昊. 涡轴发动机部件级建模及神经网络控制研究[D]. 成都: 电子科技大学, 2018.

[46] A. Assad, W. Khalaf, I. Chouaib. Radial basis function Kalman filter for attitude estimation in GPS-denied environment[J]. IET Radar, Sonar & Navigation, 2020, 14(5): 736-746.

[47] T. Tanbay. Meshless solution of the neutron diffusion equation by the RBF collocation method using optimum shape parameters[J]. Journal of Innovative Science and Engineering (JISE), 2019(1), 23-31.

[48] M. Y. Zhang, X. Y. Yu, P. X. Ding, et al. Distributed adaptive three-dimension formation control based on improved RBF neural network for non-linear multi-agent time-delay systems[J]. IET Control Theory & Applications, 2019, 13(17): 2758-2765.

[49] Y. L. Du. Forecasting hs300 based on combination of arima model and RBF neural network[C]//Proceedings of 2018 3rd BEM International Conference on Modern Education and Social Science, 2018: 46-50.

[50] 徐晴, 周超, 赵双双, 等. 基于机器学习的短期电力负荷预测方法研究[J]. 电测与仪表, 2019, 56(23): 70-75.

[51] 肖白, 刘庆永, 牛强, 等. 基于元胞负荷特性分析的RBF神经网络空间负荷预测方法[J]. 电网技术, 2018, 42(1): 301-307.

[52] H. T. Wang, L. Ye, C. Yan, et al. Study on mechanical concurrent fault diagnosis method based on PSO-MRLSSVM[C]//Proceedings of 2018 International Conference on Mathematics, Modelling, Simulation and Algorithms (MMSA 2018), 2018: 188-192.

[53] J. H. Zhang, X. Ji, F. H. Ma. Analysis of coordinated system prediction system based on least square support vector machine[C]//Proceedings of 2019 International Conference on Information Science, 2019: 663-666.

[54] 黄超. 基于最小二乘支持向量机的惯导系统精度评估方法研究[D]. 哈尔滨: 哈尔滨工业大学, 2019.

[55] S. L. Xie, G. Y. Ren. The hysteresis modeling of pneumatic artificial muscles using least squares support vector machine approach[J]. Journal of Intelligent Material Systems and Structures, 2020, 31(11): 1358-1370.

[56] S. B. Li, Y. Q. Zhou, R. Li, et al. Online lithium battery fault diagnosis based on least square support vector machine optimized by ant lion algorithm[J]. International Journal of Performability Engineering, 2020, 16(10): 1637-1645.

[57] 骆正山, 黄仁惠. 基于KPCA-APSO-LSSVM的充填管道磨损风险预测[J]. 有色金属工程, 2021, 11(3): 96-106.

[58] R. Divyesh, T. Radhakrishnan, V. Kirubakaran. Explicit nonlinear predictive control algorithms for Laguerre filter and sparse least square support vector machine-based Wiener model[J]. Transactions of the Institute of Measurement and Control, 2021, 43(4): 812-831.

[59] 向玲, 邓泽奇. 基于改进经验小波变换和最小二乘支持向量机的短期风速预测[J]. 太阳能学报, 2021, 42(2): 97-103.

[60] X. M. He, X. Jiang, X. P. Fu, et al. Least squares support vector machine regression combined with Monte Carlo simulation based on the spatial frequency domain imaging for the detection of optical properties of pear[J]. Postharvest Biology and Technology, 2018, 145: 1-9.

[61] 邓军, 雷昌奎, 曹凯, 等. 煤自燃预测的支持向量回归方法[J]. 西安科技大学学报, 2018, 38(2): 175-180.